mirror of https://github.com/saitohirga/WSJT-X.git
151 lines
4.0 KiB
Fortran
151 lines
4.0 KiB
Fortran
subroutine timf2x(k,nfft,ntrperiod,nwindow,nb,peaklimit,faclim,cx0,cx1, &
|
|
slimit,lstrong,px,nzap)
|
|
|
|
! Sequential processing of time-domain I/Q data, using Linrad-like
|
|
! "first FFT" and "first backward FFT".
|
|
|
|
! cx0 - complex input data
|
|
! nfft - length of FFTs
|
|
! nwindow - 0 for no window, 2 for sin^2 window
|
|
! cx1 - output data
|
|
|
|
! Non-windowed processing means no overlap, so kstep=nfft.
|
|
! Sin^2 window has 50% overlap, kstep=nfft/2.
|
|
|
|
! Frequencies with strong signals are identified and separated. Back
|
|
! transforms are done separately for weak and strong signals, so that
|
|
! noise blanking can be applied to the weak-signal portion. Strong and
|
|
! weak signals are finally re-combined in the time domain.
|
|
|
|
parameter (MAXFFT=32768,MAXNH=MAXFFT/2)
|
|
parameter (MAXSIGS=100)
|
|
complex cx0(0:nfft-1),cx1(0:nfft-1)
|
|
complex cx(0:MAXFFT-1),cxt(0:MAXFFT-1)
|
|
complex cxs(0:MAXFFT-1),covxs(0:MAXNH-1) !Strong X signals
|
|
complex cxw(0:MAXFFT-1),covxw(0:MAXNH-1) !Weak X signals
|
|
complex cxw2(0:8191)
|
|
complex cxs2(0:8191)
|
|
real*4 w(0:MAXFFT-1)
|
|
real*4 s(0:MAXFFT-1)
|
|
logical*1 lstrong(0:MAXFFT-1),lprev
|
|
integer ia(MAXSIGS),ib(MAXSIGS)
|
|
logical first
|
|
data first/.true./
|
|
data k0/99999999/
|
|
save
|
|
|
|
if(first) then
|
|
pi=4.0*atan(1.0)
|
|
do i=0,nfft-1
|
|
w(i)=(sin(i*pi/nfft))**2
|
|
enddo
|
|
s=0.
|
|
nh=nfft/2
|
|
nfft2=nfft/4
|
|
if(ntrperiod.ge.300) nfft2=nfft/32
|
|
nh2=nfft2/2
|
|
kstep=nfft
|
|
if(nwindow.eq.2) kstep=nh
|
|
fac=1.0/nfft
|
|
slimit=1.e30
|
|
first=.false.
|
|
endif
|
|
|
|
if(k.lt.k0) then
|
|
covxs=0.
|
|
covxw=0.
|
|
endif
|
|
k0=k
|
|
|
|
cx(0:nfft-1)=cx0
|
|
if(nwindow.eq.2) cx(0:nfft-1)=w(0:nfft-1)*cx(0:nfft-1)
|
|
call four2a(cx,nfft,1,-1,0) !First forward FFT, r2c
|
|
cxt(0:nfft-1)=cx(0:nfft-1)
|
|
|
|
! Identify frequencies with strong signals, copy frequency-domain
|
|
! data into array cs (strong) or cw (weak).
|
|
|
|
do i=0,nfft-1
|
|
s(i)=real(cxt(i))**2 + aimag(cxt(i))**2
|
|
enddo
|
|
ave=sum(s(0:nfft-1))/nfft
|
|
lstrong(0:nfft-1)=s(0:nfft-1).gt.10.0*ave
|
|
|
|
nsigs=0
|
|
lprev=.false.
|
|
iwid=1
|
|
ib=-99
|
|
do i=0,nfft-1
|
|
if(lstrong(i) .and. (.not.lprev)) then
|
|
if(nsigs.lt.MAXSIGS) nsigs=nsigs+1
|
|
ia(nsigs)=i-iwid
|
|
if(ia(nsigs).lt.0) ia(nsigs)=0
|
|
endif
|
|
if(.not.lstrong(i) .and. lprev) then
|
|
ib(nsigs)=i-1+iwid
|
|
if(ib(nsigs).gt.nfft-1) ib(nsigs)=nfft-1
|
|
endif
|
|
lprev=lstrong(i)
|
|
enddo
|
|
|
|
if(nsigs.gt.0) then
|
|
do i=1,nsigs
|
|
ja=ia(i)
|
|
jb=ib(i)
|
|
if(ja.lt.0 .or. ja.gt.nfft-1 .or. jb.lt.0 .or. jb.gt.nfft-1) then
|
|
cycle
|
|
endif
|
|
if(jb.eq.-99) jb=ja + min(2*iwid,nfft-1)
|
|
lstrong(ja:jb)=.true.
|
|
enddo
|
|
endif
|
|
|
|
do i=0,nfft-1
|
|
if(lstrong(i)) then
|
|
cxs(i)=fac*cxt(i)
|
|
cxw(i)=0.
|
|
else
|
|
cxs(i)=0.
|
|
cxw(i)=fac*cxt(i)
|
|
endif
|
|
enddo
|
|
|
|
df=12000.0/nfft
|
|
i0=nint(1500.0/df)
|
|
cxw2(0:nh2)=cxw(i0:i0+nh2)
|
|
cxw2(nfft2-nh2:nfft2-1)=cxw(i0-nh2:i0-1)
|
|
cxs2(0:nh2)=cxs(i0:i0+nh2)
|
|
cxs2(nfft2-nh2:nfft2-1)=cxs(i0-nh2:i0-1)
|
|
|
|
call four2a(cxw2,nfft2,1,1,1) !Transform weak and strong X
|
|
call four2a(cxs2,nfft2,1,1,1) !back to time domain, separately
|
|
|
|
if(nwindow.eq.2) then
|
|
cxw2(0:nh2-1)=cxw2(0:nh2-1)+covxw(0:nh2-1) !Add prev segment's 2nd half
|
|
covxw(0:nh2-1)=cxw2(nh2:nfft2-1) !Save 2nd half
|
|
cxs2(0:nh2-1)=cxs2(0:nh2-1)+covxs(0:nh2-1) !Ditto for strong signals
|
|
covxs(0:nh2-1)=cxs2(nh2:nfft2-1)
|
|
endif
|
|
|
|
! Apply noise blanking to weak data
|
|
if(nb.ne.0) then
|
|
do i=0,kstep-1
|
|
peak=abs(cxw(i))
|
|
if(peak.gt.peaklimit) then
|
|
cxw2(i)=0.
|
|
nzap=nzap+1
|
|
endif
|
|
enddo
|
|
endif
|
|
|
|
! Compute power levels from weak data only
|
|
px=0.
|
|
do i=0,kstep-1
|
|
px=px + real(cxw2(i))**2 + aimag(cxw2(i))**2
|
|
enddo
|
|
|
|
cx1(0:kstep-1)=cxw2(0:kstep-1) + cxs2(0:kstep-1) !Weak + strong
|
|
|
|
return
|
|
end subroutine timf2x
|